++i and i++

[S Senthil Kumar]

There's no advantage for integers like i. But if you're using an STL iterator, for example, there's a definite advantage using ++i over i++. The implementation of the postincrement operator would need to do whatever ++i does, but it would also need to store the original value in a temporary variable and return it, which would be less performant. Regards Senthil _____________________________ My Blog | My Articles | WinMacro

[John M Drescher]

Thanks for posting this question... I always use i++ but did not know it could be slower. John

[Owner drawn]

Recently I have started using ++i over i++. But you must be careful with assignments. For eg: int j = ++i or i++; You must be fully aware of what you are doing. I use ++i in loops. There is a slight advantage. i++ returns the value and then increments ie j = i; i = i+1; ++i simply increments ie j = (i=i + 1);

Jesus Loves You and Me :)

--Owner Drawn --Nothing special --Defeat is temporary but surrender is permanent --Never say quits --Jesus is Lord

[vikas amin]

I dont know about integeres but for objects it sure that the performance is better in ++i. U can find the reason at this link http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vccelng/htm/overl\_12.asp I did a small experiment with integers considering the loop , to test the speed but it is almost same , for bot type of increments :cool: Vikas Amin Embin Technology Bombay vikas.amin@embin.com

[Nishad S]

Thank you. - NS -

[Nishad S]

Thank you. - NS -

[Nishad S]

I was also using i++. Recently changed to ++i... :) Thank you. - NS -

[Nishad S]

Thank you. - NS -

[G Haranadh]

YoU ArE MoSt WeLcOmE. ;) Nice talking to you. :-O
If you judge people, you have no time to love them. -- Mother Teresa

[S Senthil Kumar]

G Haranadh wrote:

++i is faster than i++.

Could you explain why? It might be a 100th of a ms faster, but other than that, I don't see why ++i would be significantly faster than i++. Regards Senthil _____________________________ My Blog | My Articles | WinMacro

[Prakash Nadar]

from the VC assembly dump.

10: for(i=0;i<5; i++)
00401028 mov dword ptr [ebp-4],0
0040102F jmp main+2Ah (0040103a)
00401031 mov eax,dword ptr [ebp-4]
00401034 add eax,1
00401037 mov dword ptr [ebp-4],eax
0040103A cmp dword ptr [ebp-4],5
0040103E jge main+3Fh (0040104f)
11: printf("Hello World!\n");
00401040 push offset string "Hello World!\n" (0042001c)
00401045 call printf (004010b0)
0040104A add esp,4
0040104D jmp main+21h (00401031)
12:
13: for(i=0;i<5; ++i)
0040104F mov dword ptr [ebp-4],0
00401056 jmp main+51h (00401061)
00401058 mov ecx,dword ptr [ebp-4]
0040105B add ecx,1
0040105E mov dword ptr [ebp-4],ecx
00401061 cmp dword ptr [ebp-4],5
00401065 jge main+66h (00401076)
14: printf("Hello World!\n");
00401067 push offset string "Hello World!\n" (0042001c)
0040106C call printf (004010b0)
00401071 add esp,4
00401074 jmp main+48h (00401058)

so clearly the code for both cases is the same. so no difference if you use the incrementor statements (pre or post) that are independent. -Prakash -Prakash

[Nishad S]

Its amazing... I thought that post increment will cause a temporary instance for keeping the value, and then it will increment. Thank you for revealing a great mistake. - NS -

[vikas amin]

As far as i know it will not effect the integer type but if u are using an c++ object during the increment it will effect preformance Vikas Amin Embin Technology Bombay vikas.amin@embin.com

[vikas amin]

Now i foudnd something that can prove ++i is better then i++. this is in the case ur passing a argument to a function in my case the fun name is add. So ++i i better as it donot create any temp variable to store . As the value is need to be incremented first . But in the late case the value have to be incremented afterwards so a temp variable is needed.. Assmebly code to prove my words U r wrong the case is vice versa Tough to Belive Assembly code to prove my observations 507: Add(++i); 004010CE mov eax,dword ptr [ebp-4] 004010D1 add eax,1 004010D4 mov dword ptr [ebp-4],eax 004010D7 mov ecx,dword ptr [ebp-4] 004010DA push ecx 004010DB call @ILT+15(Add) (00401014) 004010E0 add esp,4 508: Add(i+1); 004010E3 mov edx,dword ptr [ebp-4] 004010E6 add edx,1 004010E9 push edx 004010EA call @ILT+15(Add) (00401014) 004010EF add esp,4 506: Add(1+i); 004010CE mov eax,dword ptr [ebp-4] 004010D1 add eax,1 004010D4 push eax 004010D5 call @ILT+15(Add) (00401014) 004010DA add esp,4 507: Add(i++); 004010DD mov ecx,dword ptr [ebp-4] 004010E0 mov dword ptr [ebp-5Ch],ecx 004010E3 mov edx,dword ptr [ebp-5Ch] 004010E6 push edx 004010E7 mov eax,dword ptr [ebp-4] 004010EA add eax,1 004010ED mov dword ptr [ebp-4],eax 004010F0 call @ILT+15(Add) (00401014) 004010F5 add esp,4 I dont know but how i+1 is more effecient to pass then i++ or ++i Vikas Amin Embin Technology Bombay vikas.amin@embin.com

[segment_fault]

I'd reiterate the earlier comment that you should BE VERY CAREFUL mixing ++i and i++. It's very easy to skip index 0 in situations where you pre-increment.. only to cause bugs that are sometimes subtle -- especially when dealing with interating through arrays of data. Read more of what I have to say at http://directx9.blogspot.com/